Continuously identifying users via interaction
2013-11-22 16:25:39.632746+00 by
Dan Lyke
2 comments
Back in the late 1980s I tossed this idea around while I was in college, and had it scoffed at by various faculty members. Part of why I decided to drop out.
Users ID'ed through typing, mouse movements:
A series of 90 minute typing tests carried out on 2000 people at Iowa State University found users could be identified with a half percent margin of error based on the way they hit keys.
Capturing Keystroke Dynamics for Active User Authentication (PDF). Via /..
[ related topics:
Current Events Education
]
comments in ascending chronological order (reverse):
#Comment Re: made: 2013-11-23 13:02:12.933948+00 by:
DaveP
Not positive, but I strongly suspect you would need multiple profiles. Hungover Dave doesn't type
the same as Sleepy Dave, who doesn't type the same as well-rested Dave, who doesn't type the
same as over-caffeinated Dave.
#Comment Re: made: 2013-11-26 18:49:46.499414+00 by:
Dan Lyke
Got to thinking about this, and realized that one of the problems with customizable interfaces is that when someone new takes overmy computer for a short time, they get completely confused. What if my EDITOR
environment variable could tell if I'm the current active user and give me Emacs with my ~/.emacs.d/init.el
, but offer Charlene gedit?
What if, when we're driving and someone calls or I want to use the phone for nav and I hand my phone to her and say "can you...", it gives her something more approximating the default, or even her interface, than my customized one?